I am currently working on the foundations
of machine learning and the application of evolutionary
algorithms to Real World learning problems. The search
for foundations of machine learning leads to three main questions:
How should a learning system represent and process uncertain
information, or, what is the proper inductive logic?
What set of possible models should the system consider?
How to relate the evidence for a
model to its complexity?
In the long run, a general learning system
should be able to detect as many regularities in its percept stream as
possible, while dealing sensibly with the inherent uncertainty of predictions
based on a finite amount of data.
Participants of the WNA 2011-Workshop (from left to right): Yupeng Cun, Ashutosh
Malhotra, Paurush Praveen, Mikael Gast, Mufassra Naz, Seraya Maouche, Steve
Horvath, Khalid Abnaof, Katrin Illner, Jörg Zimmermann.